[EMAIL PROTECTED] said:
> I have read all FAQs, how-to etc., but my incomming messages are still marked
> as "autolearn=no" Any other idea, please ?

autolearn=no does not mean that Bayes is not working, it just means that the 
scores were such that it decided not to learn from the mail. (This is in the 
FAQ BTW).

Your Bayes DB needs to have either been fed, or auto-learned at least 200 ham 
and 200 spam for it to work.

Try:

sa-learn --dump magic

To check these figures.

Here's my output:

> sa-learn --dump magic
0.000          0          2          0  non-token data: bayes db version
0.000          0      51071          0  non-token data: nspam
0.000          0      40583          0  non-token data: nham
0.000          0     139618          0  non-token data: ntokens
0.000          0 1088319955          0  non-token data: oldest atime
0.000          0 1088691427          0  non-token data: newest atime
0.000          0 1088688835          0  non-token data: last journal sync atime
0.000          0 1088579369          0  non-token data: last expiry atime
0.000          0     259198          0  non-token data: last expire atime delta
0.000          0      41984          0  non-token data: last expire reduction 
count

So Bayes has learnt from 40583 hams and 51071 spams.

Owen

--
 Via Net.Works UK Ltd
 Local Touch Global Reach 
 Owen McShane                   Systems Administrator
 http://www.vianetworks.co.uk   Tel +44 (0)1925 484444

Reply via email to